E-Mail [email protected]
Original Paper
Intervirology 2013;56:217–223 DOI: 10.1159/000348511
Discrepancies of HIV-1 Reverse Transcriptase Resistance Interpretation of Insertions and Deletions between Two Genotypic Algorithms
Giselle Ibette Silva López-Lopes André Minhoto Lança
João Leandro de Paula Ferreira Luciana Oliveira Souza
Luís Fernando de Macedo Brígido
Laboratório de Retrovírus, Centro de Virologia, Instituto Adolfo Lutz, São Paulo , Brazil
drugs, predominantly for TDF, d4T and ETV, when sequences with deletions were evaluated. Conclusion: Both indel posi-tioning and its impact on drug susceptibility varies depend-ing on the algorithm, a fact that might influence the clinical decision. Critical analysis of indel sequences with manual alignments is important, and its use alongside different algo-rithms may be important to better understand the outcomes of genotypic resistance prediction.
Copyright © 2013 S. Karger AG, Basel
Introduction
The HIV-1 polymerase gene is responsible for coding different enzymes which are essential for the retrovirus replicative cycle, including the protease and reverse tran-scriptase (RT) that are targets for most of HIV-1 antiret-roviral (ARV) drugs. During reverse transcription, a lack of the proofreading properties of the enzyme, known for its relatively high error rate, is one of the factors respon-sible for viral diversity, an exquisite environment for the emergence of mutant viral strains. Also, the selective ef-
Key Words
Highly active antiretroviral therapy · HIV-1 antiretroviral drugs · HIV Stanford Resistance Database · Geno2Pheno [resistance]
Abstract
Background: Bioinformatics algorithms have been devel-oped for the interpretation of resistance from sequence sub-mission, which supports clinical decision making. This study evaluated divergences of the interpretation of the genotyp-ing in two commonly used algorithms, using sequences with indels of reverse transcriptase genes. Methods: Sequences were obtained from virus RNA of patients failing highly active antiretroviral therapy from 2004 to 2011. Alignments were obtained using Clustal W including subtype B consensus and HXB2. Sequences with evidence of indels were submitted to the Stanford Resistance Database and to the Geno2Pheno to locate indel positioning and determine the resistance pro-file. Results: A total of 1,959 partial reverse transcriptase se-quences were assessed, mostly subtype B (74%). Insertions and deletions were observed in 0.9 and 0.6% of sequences, respectively. Discordant insert positioning was assigned for most (90%) insertion sequences, with 27% discordances for deletions. Susceptibility differed for some antiretroviral
Received: August 14, 2012 Accepted after revision: January 22, 2013 Published online: May 9, 2013
Dr. Luís Fernando de Macedo Brígido, MD PhD Dr. Arnaldo Av., 355 Cerqueira Cesar São Paulo 01246-902 (Brazil) E-Mail lubrigido @ gmail.com
© 2013 S. Karger AG, Basel0300–5526/13/0564–0217$38.00/0
www.karger.com/int
G.I.S. López-Lopes and A.M. Lança contributed equally to this work.
Dow
nloa
ded
by:
Coo
rden
ador
ia d
e C
ontr
ole
de D
oenc
as d
a S
ecre
taria
de
20
0.14
4.1.
138
- 6/
24/2
014
7:09
:10
PM
López-Lopes/Lança/de Paula Ferreira/Souza/de Macedo Brígido
Intervirology 2013;56:217–223DOI: 10.1159/000348511
218
fect of the several ARV drugs composing the armamen-tarium of HIV-1 therapy on viral replication is essential to the development of the resistance observed in treat-ment failure [1] . As mutations can be transmitted [2] , its impact on treatment options, both for initial and salvage therapies, constitutes an important public health issue. Over two hundred mutations have been associated with drug resistance, with complex mutation patterns in pa-tients on highly active antiretroviral therapy (HAART) [3] . The evaluation of susceptibility to ARV drugs is of great importance in providing alternatives to salvage therapy in patients failing treatment, and genotyping tests have been employed to identify mutations that may confer resistance to ARV drugs [4] . The majority of sub-stitutions confer cross-resistance drugs of the same class [5] , such as K65R that reduces susceptibility to almost all nucleotide analogue reverse transcriptase inhibitors (NRTI) [6] . Moreover, strains exhibiting insertions of multiple nucleotides in protease and RT genes have been found in patients who have failed ARV therapy, and compensatory mutations are favored by drug-selective pressure on resistant strains replicating in a viremic pa-tient [7] . Among the various mutations that confer re-duced susceptibility, insertions and deletions, often called indels, can be observed in 0.5–2.4% of heavily treated patients [8, 9] .
Along with sequence data per se , different interpreta-tion algorithms have been developed to allow fast and concise resistance profile definition from sequence sub-mission. Some of the genotyping algorithms are freely available, namely the HIV Stanford Resistance Database (HIVdb) [5] and the Geno2Pheno [resistance] (G2P) [10] ; these are two of the algorithms most commonly used by researchers and physicians worldwide. However, while most of the time these various systems agree, the inter-pretation of some sequences can be rather divergent. Several authors have made use of different criteria to es-tablish a comparison between those different methods of resistance prediction [11–18] . However, the literature lacks a comparison of algorithms regarding when inser-tions or deletions of amino acids are present and how the positioning of these codons in the websites contrib-utes to different susceptibility predictions. The aim of this study was to evaluate the divergences in the inter-pretation of the genotyping in two different algorithms using a routine resistance sequence dataset with inser-tions and deletions in the RT gene. Also, the association of the indels with other mutations was evaluated to bet-ter understand their role in patients extensively exposed to HAART.
Patients and Methods
Sequence Selection Sequences were obtained from plasma RNA of patients failing
HAART, with a viral load over 2,000 copies/ml (3.3 log 10 ), while they were attending facilities administered by the public health system of São Paulo and nearby cities, as part of routine resistance genotyping from 2004 and 2011. HIV-1 RT gene sequences were obtained using ViroSeq (Abbott Laboratories, USA) resolved on ABI 3100 Genetic Analyzer (until 2008) or TRUGENE ® HIV-1 Genotyping Kit for Drug Resistance resolved on OpenGene (Sie-mens Healthcare Diagnostics, USA) thereafter. All sequences ob-tained were screened for the presence of insertions and deletions.
Multiple sequence alignments were obtained using the Clustal W algorithm embedded in BioEdit software including subtype B consensus and HXB2 reference sequences downloaded from Los Alamos National Laboratory (LANL) database (http://www.hiv.lanl.gov/content/sequence/HIV/mainpage.html) [19] .
Sequence Submission and Indel Positioning Sequences with evidence of indels were submitted to HIVdb
version 6.2.0 (last update May 29, 2012) and G2P version 3.3 (last update September 13, 2012) to detect and locate insertion/deletion positioning, and to determine the drug susceptibility profile. Man-ual entering of individual mutations in HIVdb was also evaluated to verify if this would change the prediction made by HIVdb auto-matic submission. Moreover, a manual alignment was conducted to allow indel positioning. The determination of which would be the most probable positioning of the indels was based on the prior conservative assumption that deletions at position 67 and inser-tions at codon 69 are the most common changes in this genomic region associated with genetic resistance to ARV drugs [20, 21] . The amino acid composition of all sequences was evaluated to doc-ument the prevalence of resistance mutations in both groups.
Interpretation of Genotyping HIVdb stratifies resistance in five levels (H = high-level resis-
tance, I = intermediate resistance, L = low-level resistance, P = po-tential low-level resistance and S = susceptible), while G2P strati-fies it in three levels (1 = resistant, 2 = intermediate and 3 = sus-ceptible). Our paired analysis collapsed the scores ‘potential low-level resistance’ as ‘susceptible’ and the low level as ‘intermedi-ate resistance’ (1 = S, P, L; 2 = L, I; 3 = H), as suggested by the HIVdb HIValg program. Differences between algorithms were evaluated as partial discordance, when the algorithms disagree at the level of resistance or between susceptible and intermediate, or as full dis-cordance, when one algorithm scores a sequence as susceptible and the other shows a fully resistant profile. The International AIDS Society mutation list update from 2011 [5] was also used in the characterization of mutations associated with resistance.
Results
A total of 2,272 RT partial gene sequences were screened and those with available data such as age, gen-der, CD4, viral load and treatment regimens were select-ed. Also, only the first sequences from each patient were
Dow
nloa
ded
by:
Coo
rden
ador
ia d
e C
ontr
ole
de D
oenc
as d
a S
ecre
taria
de
20
0.14
4.1.
138
- 6/
24/2
014
7:09
:10
PM
Discrepancies in Interpretation of RT HIV-1 Insertions and Deletions
Intervirology 2013;56:217–223DOI: 10.1159/000348511
219
included, i.e. a total of 1,959 sequences. Thirty (1.4%) of these were found to contain amino acid insertions (0.9%) or deletions (0.6%) ( fig. 1 ). Patients’ characteristics ac-cording to the indel type as well as from the remaining 1,929 of nonindel cases are detailed in table 1 .
Sequence Analysis Nineteen sequences exhibited an insertion of 1, 2 or 8
amino acids at codon 69 according to manual alignment ( fig. 1 ). Thirteen presented the usual insertion of 2 amino acids (69ins 2 ) followed by T69S substitution in most of those sequences (92.3%). The insertions observed were seven SG (53%), three SS (23.6%), two VT (15.4%) and one CG (7.7%). Five sequences showed the presence of 1-amino acid insertion (69ins 1 ), mostly T (60%), and all those sequences exhibited the substitution K64N. Addi-tionally, one sequence showed an unusual 8-amino acid insertion (69ins 8 ). Along the insertion sequences eleven isolates had deletions at codon 67 (Δ67); the substitution T69G was observed in all strains and the mutations S68N or S68D in 16.2% each.
Sequences were also evaluated for resistance profiles according to the 2011 IAS [5] resistance mutation list grouped by Δ67, 69ins 1 , 69ins 2 and 69ins 8 . Within the 69ins 2 group, the prevalent resistance mutations were the thymidine analogue mutations type I (TAM-I), as well as
A62V and M184 V/I. On the other hand, along the 69ins 1 , the most common resistance mutations were the TAM type II (TAM-II) followed by T215F. The third group, Δ67, had a high prevalence of both TAM-I and TAM-II, as well as L74I and M184V. The single 69ins 8 sequence showed the TAM-I substitutions M41L, L210W and T215F, as well as the TAM-II K70R. Data of the NRTI mutations are detailed in table 2 .
Comparison of Algorithms HIVdb and G2P alignment algorithms were compared
both as positioning concordance as well as indel impact on drug susceptibility. Regarding positioning, the insertions were discordant for 90% of sequences, while the discor-dance was 27% in samples with deletions. All 19 events of insertion were positioned at codon 69 by HIVdb, while positioning varied between 64 and 69 when using G2P. Sequences defined by HIVdb as 69SG/SS were alternative-ly classified as 66ins 2 in G2P, with the exception of one 69SS sequence defined as 67ins 2 . Those containing 69CG/VT had its positioning set as 68ins 2 in G2P. No difference was observed among the 69D and 69N sequences, both be-ing classified as 69ins 1 . The three 69T insertions were po-sitioned at 68 by G2P. Furthermore, the sequence contain-ing the 8-amino acid insertion was positioned at codon 64. The patient with this variant (BR067SP07) had another
Fig. 1. Deduced amino acids of the indel regions with positioning according to the manual alignment, to G2P and to HIVdb. Hyphens represent gaps in the align-ment, while black boxes highlight the dif-ferent G2P and HIVdb positioning of in-dels. Codon numbering is adjusted to the HXB2 reference. Samples BR185SP11 * and BR577SP09 * are subsequent samples of patients BR085SP10 and BR067SP07, respectively.
Colo
r ver
sion
ava
ilabl
e on
line
Insertions
Deletions
HIVdbGeno2phenoManual alignment
65 70 65 70 65 70
Dow
nloa
ded
by:
Coo
rden
ador
ia d
e C
ontr
ole
de D
oenc
as d
a S
ecre
taria
de
20
0.14
4.1.
138
- 6/
24/2
014
7:09
:10
PM
López-Lopes/Lança/de Paula Ferreira/Souza/de Macedo Brígido
Intervirology 2013;56:217–223DOI: 10.1159/000348511
220
genotype test performed 2 years later that showed the per-sistence of this insertion, as well one other case with sub-sequent sample evaluation; these are included in figure 1 . Regarding the deletions, all 7 sequences exhibiting the wild-type S68 (according to manual alignment) were clas-sified as Δ67 in both HIVdb and G2P, while the two S68N mutations had their deletion positioned at codon 69 by HIVdb and 67 by G2P. Moreover, the remaining two se-quences exhibiting S68D were classified as Δ70 by G2P, while HIVdb positioned the deletion at 68 and 69 codons ( fig. 1 ). The manual entering of mutations either by typing
or by using the drop-down menu offered as HIVdb Muta-tion List option did not differ from the susceptibility re-sults at automatic sequence analysis. However, some dif-ferences were seen at positioning, when the drop-down menu was used, where some uncommon changes exhib-ited at automatic analysis were not available for selection, e.g. BR11SP151’s K70d (deletion). This could be solved by manually typing in those mutations. Both alternatives did not change the susceptibility profile.
The prediction of susceptibility to each drug of the two algorithms was evaluated separately for sequences har-
Table 1. Clinical and demographic characteristics of the study patients according to indel type
Demographic/clinical variables Nonindel (n = 1,929) 67Δ (n = 11) 69ins1 (n = 5) 69ins2 (n = 13) 69ins8 (n = 1)
Age, years 38 (30–45) 41 (19–42) 14 (11–17) 41 (17–46) 42Gender (male) 60 71.4 40 92.3 0CD4+ T cells, cells/mm3 238 (117–399) 146 (28–465) 430 (155–553) 300 (112–389) 381Nadir CD4+ T cells, cells/mm3 120 (47–233) 38 (6–292) 428 (121–496) 149 (98–267) 381Viral load (log10) 4.49 (4.01–4.98) 4.6 (3.8–4.7) 4.65 (4.33–5.78) 4.65 (4.27–5.02) 5.6Exposure, weeks 348 (227–457) 409 (388–668) 379 (324–430) 398 (345–521) 168Number of treatments 3 (2–4) 4.5 (2.5–5.0) 1 (1–2) 3 (2–4.5) 1Initial therapy
Mono 6.9 0 0 7.7 0Double 34.7 54.5 100 46.2 0HAART 58.4 45.5 0 46.2 100
Present therapyMono 0.1 0 0 0 0Double 3.4 0 60 0 0HAART 96.5 100 40 100 100
SubtypeB 73.7 81.8 40 61.2 100F 7.4 9.1 – 23.1 –C 2.6 9.1 – 7.7 –Recombinant mosaic 15.8 – 60 7.7 –Unknown 0.5 – – – –
Results are expressed as the median with the interquartile range in parentheses or as percentage. Nonindels = From sequences with-out indels at RT; Δ67 = deletion at codon 67; 69ins1, 69ins2 and 69ins8 = insertion of 1, 2 and 8 amino acids at codon 69, respectively.
Table 2. Resistance mutations associated with NRTIs in the different groups of indels according to the ‘IAS Resistance Update 2011’ [5]
n M41L A62V K65R D67N K70R L74I/V V75I/M F77L Y115F F116Y Q151M M184I/V L210W T215F/Y K219E/Q
Control 1,929 33 4 2 30 20 4/4 2/7 3 1 2 2 1/62 24 12/37Δ67 11 36 9 73 45/0 9 9 9 0/91 18 46/36 80/069ins1 5 20 40 100 0/20 0/20 40/0 0/10069ins2 13 69 46 12/12 0/31 15 23/54 54 17/75 0/869ins8 1 100 100 100 100/0
Results are expressed as percentage. TAM-I are given in bold, while TAM-II are given in italics.
Dow
nloa
ded
by:
Coo
rden
ador
ia d
e C
ontr
ole
de D
oenc
as d
a S
ecre
taria
de
20
0.14
4.1.
138
- 6/
24/2
014
7:09
:10
PM
Discrepancies in Interpretation of RT HIV-1 Insertions and Deletions
Intervirology 2013;56:217–223DOI: 10.1159/000348511
221
boring insertions or deletions. The exhibited partial and full discordance between algorithms is detailed in table 3 , with some discordance observed for all ARV drugs. Par-tial or full discordance was observed for a new generation of non-NRTI drugs, ETV (37% for insertions and 73% for deletions), as for NRTI drugs, especially d4T and TDF. Additionally, there was full discordance in 2 (10%) inser-tion samples (ddI and ETR). Full discordance was more common among sequences with deletions (36%). Using different combinations of HIVdb classifications (e.g. con-sidering low as susceptible or potential-low as intermedi-ate at G2P) did not show important differences in algo-rithm comparisons. It is of note that all full discordant sequences were resistant at HIVdb and susceptible at G2P.
Discussion
Several computational algorithms of the interpretation of the genotyping of HIV-1 viral strains have been devel-oped to better understand and characterize resistance in clinical samples. Alongside patients’ treatment history, the genotypic algorithms may be employed to predict re-sistance to ARV drugs and help physicians with a fast and reliable method of determining the most adequate HAART option for each individualized salvage regimen. However, differences between the construction and the in vivo and in vitro resistance information used for each algorithm development may have an impact on phenotypic predic-tion and may lead to different resistance profiles. Two of the most frequently used algorithms available, HIVdb and G2P, were selected for comparison. Our cohort includes mostly heavily treated patients, with a median of 7 years on ARV treatment, in many of whom treatment was initi-ated in the 1990s. From all sequences screened for indels, deletions (0.6%) and insertions (0.9%) among patients failing HAART were present in a prevalence that did not differ from the findings of Tamalet et al. [8] , Winters and Merigan [7] and Harrigan et al. [9] .
The indel groups, separated by the number of amino acids inserted, and the nonindel (control) group were compared. With exception of 69ins 1 , these groups were highly homogeneous regarding demographic, laborato-ry and treatment data ( table 1 ). However, the 69ins 1 group was mostly comprised of children (80% under 18 years old), a population with immunological, viro-logical and therapeutic peculiarities (e.g. ARV options available to children and adolescents were more limited at that time, and dual NRTI therapy was used for longer periods of time). Among insertion sequences, a high het-
erogeneity of both substitutions and insertions was ob-served, where 1, 2 or 8 amino acids were identified. Also, the genotypic resistances to NRTI differed among the sequences harboring one and two insertions. The resis-tance mutations exhibited by 69ins 2 sequences were in agreement with the IAS 2011 mutation list regarding the 69 insertion complex [5] . Interestingly, recently HIVdb [22] has included in the 69 insertion comments as sepa-rate observations regarding 69 single and double inserts, where the former are often accompanied by TAM-II, while the latter are followed by TAM-I. Our data agree with HIVdb and suggest that a distinct mutation pattern may be associated with either a single or a double inser-tion, with potential impact on drug susceptibility. K64N, a rare mutation found by Rakik et al. [23] on a 69ins 1 sequence, was also observed on all our sequences har-boring single insertions. A distinct insertion of 8 amino acids in this same region (69ins 8 ) was detected in one case. A second sample, obtained 2 years after the first genotyping and a subsequent change of drug therapy, confirmed the presence of the long insertion, along some substitutions in the nearby positions ( fig. 1 ). Large inser-tions between codons 64 and 70 are rather uncommon, although insertions/duplications of up to 15 amino acids have already been described [9, 24] . Furthermore, a very similar substitution to the one observed in our patient was found by Van der Hoek et al. [25] , who suggested
Table 3. NRTI and NNRTI drugs susceptibility algorithms pre-diction agreement between Stanford HIVdb (version 6.2.0) and Geno2Pheno (version 3.3)
ARV Stanford HIVdb and geno2pheno discordances
inserti ons (n = 19) deletions (n = 11)
partial full partial full
ZDV 0% 0% 18% 0%ddI 16% 5% 0% 0%d4T 37% 0% 36% 27%3TC 32% 0% 9% 0%ABC 5% 0% 27% 0%TDF 26% 0% 73% 0%NVP 11% 0% 0% 9%EFV 21% 0% 36% 9%ETR 32% 5% 73% 0%
Partial = partial discordant; full = full discordant. Partial dis-cordance was considered when the level of resistance disagreed, as one algorithm classified as intermediate resistance, while the other classified as resistant or susceptible. Full discordant refers to full disagreement, as susceptible versus fully resistant.
Dow
nloa
ded
by:
Coo
rden
ador
ia d
e C
ontr
ole
de D
oenc
as d
a S
ecre
taria
de
20
0.14
4.1.
138
- 6/
24/2
014
7:09
:10
PM
López-Lopes/Lança/de Paula Ferreira/Souza/de Macedo Brígido
Intervirology 2013;56:217–223DOI: 10.1159/000348511
222
that this large insertion would be the result of both the duplication of the virus nucleotides and the insertion of fifteen nucleotides from the human genomic DNA. No sequence of our patient’s virus prior to HAART was available to investigate if the insertion emerged during or prior to therapy.
Susceptibility to ARV drugs is the most important as-pect of resistance algorithm interpretation. To evaluate the impact of indels, both HIVdb and G2P were com-pared to identify interpretation mismatches for each dif-ferent drug. We found a low algorithm prediction agree-ment for many drugs, both NRTI and non-NRTI in se-quences harboring insertions, mainly to d4T, ETR, TDF and 3TC. An even higher degree of discordance was ob-served in Δ67 sequences for d4T, TDF, EFV and ETR. It is of note that a quarter of the Δ67 sequences were con-sidered susceptible to d4T by G2P, while they were con-sidered fully resistant according to HIVdb. The differ-ences between indel positioning and susceptibility to ARV in both HIVdb and G2P reveal a lack of homogene-ity in the way in which these algorithms handle the dif-ferent indel strains. In fact, the HIVdb algorithm consid-ers the mutation individually and by default treats any insertion or deletion between positions 66 and 71 as Δ67 and 69ins, respectively, and the impact of each individ-ual mutations is unclear. G2P, although identifies the indels, does not elucidate how each mutation is taken into account in determining the degree of resistance. We
artificially changed the sequence by creating or remov-ing an indel, and observed the way in which the algo-rithm interprets other mutations in the same sequence changes, such as T215Y/F, by scoring from polymor-phism to resistance mutation (data not shown). These differences in the alignment and the interpretation of the genotyping of both algorithms can be fundamental in explaining why a high divergence between algorithms was observed for each drug. Therefore, in the absence of a phenotypic resistance test, a critical analysis of the se-quence, along with the treatment history of the patient might be required for the proper interpretation of resis-tance prediction, especially in the presence of indels. The use of different algorithms simultaneously, along with a manual alignment of the sequences, might help identify what could be the most probable correct position of the change, and whether it corroborates or not with the au-tomatic prediction available at those websites.
Acknowledgments
We would like to thank Fabio A.B. Cabral, Jaqueline S. Cav-alcanti, Joao P.G. Batista, Antonio F.A.C. Siqueira, Karina P. Bar-roso and Mariucha C. Barbosa for part of sample analysis and the staff of the Retrovirus Laboratory, Adolfo Lutz In stitute – Sao Paulo, that have greatly improved this work. The Brazilian AIDS National Program provided the resistance tests. Funding was received from FAPESP 2006/61311-0 and 2011/21958-2.
References
1 Hirsch MS, Brun-Vezinet F, Clotet B, et al: Antiretroviral drug resistance testing in adults infected with human immunodeficien-cy virus type 1: recommendations of an Inter-national AIDS Society-USA Panel. Clin Infect Dis 2003; 37: 113–128.
2 Wainberg MA, Cameron DW: HIV resistance to antiviral drugs: public health implications. Drug Resistance Updates 1998; 1: 104–108.
3 Larder B, Wang D, Revell A: Application of artificial neural networks for decision support in medicine. HIV Resistance Response Data-base Initiative, London, UK. Methods Mol Biol 2008; 58: 123–136.
4 Hertogs K, de Béthune MP, Miller V, et al: A ra-p id method for simultaneous detection of ph-enotypic resistance to inhibitors of protease and reverse transcriptase in recombinant human immunodeficiency virus type 1 isolates from patients treated with antiretroviral drugs. Anti-microb Agents Chemother 1998; 42: 269–276.
5 Johnson VA, Calvez V, Günthard HF, et al: 2011 update of the drug resistance mutations in HIV-1. Top Antivir Med 2011; 19: 156–164.
6 Deval J, White KL, Miller MD, et al: Mecha-nistic basis for reduced viral and enzymatic fitness of HIV-1 reverse transcriptase con-taining both K65R and M184V mutations. J Biol Chem 2004; 279: 509–516.
7 Winters MA, Merigan TC: Insertions in the human immunodeficiency virus type 1 prote-ase and reverse transcriptase genes: clinical impact and molecular mechanisms. Antimi-crob Agents Chemother 2005; 49: 2575–2582.
8 Tamalet C, Yahi N, Tourrès C, et al: Multi-drug resistance genotypes (insertions in the β3-β4 finger subdomain and mdr mutations) of HIV-1 reverse transcriptase from exten-sively treated patients: incidence and associa-tion with other resistance mutations. Virolo-gy 2000; 270: 310–316.
9 Harrigan PR, Mo T, Hirsch J, et al: A 21-base pair insertion/duplication at codon 69 of the HIV type 1 reverse transcriptase in a pa-tient undergoing multiple nucleoside therapy. AIDS Res Hum Retroviruses 2007; 24: 895–899.
10 Beerenwinkel N, Däumer M, Oett M, et al: Geno2pheno: estimating phenotyping drug resistance from HIV-1 genotypes. Nucleic Acids Res 2003; 31: 3850–3855.
11 Stürmer M, Doerr HW, Staszewski S, et al: Comparison of nine resistance interpretation systems for HIV-1 genotyping. Antivir Ther 2003; 8: 239–244.
12 De Luca A, Cingolani A, Di Giambenedetto, et al: Variable prediction of antiretroviral treat-ment outcome by different systems for inter-preting genotypic human immunodeficiency virus type 1 drug resistance. J Infect Dis 2003; 187: 1934–1943.
13 Ravela J, Betts BJ, Brun-Vezinet F, et al: HIV-1 protease and reverse transcriptase mutation pat terns responsible for discordances be-tween genotypic drug resistance interpreta-tion algorith ms. J Acquir Immune Defic Syn-dr 2003; 33: 8–14.
14 Kijak GH, Rubio A, Pampuro SE, et al: Dis-crepant results the interpretation of HIV-1 drug-resistance genotypic data among widely used algorithms. HIV Med 2003; 4: 72–78.
Dow
nloa
ded
by:
Coo
rden
ador
ia d
e C
ontr
ole
de D
oenc
as d
a S
ecre
taria
de
20
0.14
4.1.
138
- 6/
24/2
014
7:09
:10
PM
Discrepancies in Interpretation of RT HIV-1 Insertions and Deletions
Intervirology 2013;56:217–223DOI: 10.1159/000348511
223
15 Snoeck J, Kantor R, Shafer RW, et al: Discor-dances between interpretation algorithms for genotypic resistance to protease and reverse transcriptase inhibitors of human immuno-deficiency virus are subtype dependent. Anti-microb Agents Chemother 2006; 50: 694–701.
16 Liu L, May S, Richman DD, et al: Comparison of algorithms that interpret genotypic HIV-1 drug resistance to determine the prevalence of transmitted drug resistance. AIDS 2008; 22: 835–839.
17 Ntemgwa M, Gill MJ, Brenner BG, et al: Dis-crepancies in assignment of subtype/recom-binant forms by genotyping programs for HIV type 1 drug resistance testing may falsely predict superinfection. AIDS Res Hum Retro-viruses 2008; 24: 995–1002.
18 Kandathil AJ, Kannangai R, Cherian O, et al: A comparison of interpretation by three dif-ferent HIV type 1 genotyping drug resistance algorithms using sequences from non-clade B HIV type 1 strains. AIDS Res Hum Retrovi-ruses 2009; 25: 315–318.
19 Sequence search interface. HIV sequence da-tabase. Los Alamos National Laboratory (last GenBank update: June 18, 2012; accessed: June 31, 2012). http://www.hiv.lanl.gov/con tent/sequence/HIV/mainpage.html.
20 Ross L, Johnson M, Ferri RG, et al: Deletions in the β3-β4 hairpin loop of HIV-1 reverse transcriptase are observed in HIV-1 isolated from subjects during long-term antiretroviral therapy. J Hum Virol 2000; 3: 144–149.
21 Imamichi T, Murphy MA, Imamichi H, Lane HC: Amino acid deletion at codon 67 and Thr-to-Gly change at codon 69 of human im-munodeficiency virus type 1 reverse tran-scriptase confer novel drug resistance pro-files. J Virol 2001; 75: 3988–3992.
22 HIVdb Program: Genotypic Resistance Inter-pretation Algorithm. HIV Drug Resistance Database. Stanford University (update: May 29, 2012; accessed: in June 31, 2012). http://hivdb.stanford.edu.
23 Rakik AR, Ait-Khaled M, Griffin P, et al: A novel genotype encoding a single amino acid insertion and five other substitutions between residues 64 and 74 of the HIV-1 reverse tran-scriptase confers high-level cross-resistance to nucleoside reverse transcriptase inhibitors. Abacavir CNA2007 International Study Gr-oup. J Acquir Immune Deficit Syndr 1999; 22: 132–139.
24 Lobato RL, Kim EY, Kagan RM, Merigan TC: Genotypic and phenotypic analysis of a novel 15-base insertion occurring between codons 69 and 70 of HIV type 1 reverse transcriptase. AIDS Res Hum Retroviruses 2002; 18: 733–736.
25 Van der Hoek L, Back N, Jebbink MF, et al: Increase multinucleoside drug resistance and decreased replicative capacity of a hu-man immunodeficiency virus type 1 vari-ant with an 8-amino-acid insert in the re-verse transcriptase. J Virol 2005; 79: 3536–3543.
Dow
nloa
ded
by:
Coo
rden
ador
ia d
e C
ontr
ole
de D
oenc
as d
a S
ecre
taria
de
20
0.14
4.1.
138
- 6/
24/2
014
7:09
:10
PM
